Are you familiar with the concept of "huge data"? How much would it take in terms of resources to crawl and index all websites available out there? Do you know that a project like ChatGPT needs insane GPU power to run only with the limited dataset for training the model?
Have you ever wondered why so many engineers work within Google? The main technical issue of Google is to create new software to run on the CURRENT day technology, not to make a neuro and linguistic model that needs unlimited resources.
Is GPT-3 technology impressive? - Yes, it is.
Is Google already working on similar technology? - Yes, it is.
Will GPT-3, GPT-4, or any alternative algorithm ever be used to improve Google? Most likely, yes. Google is already invested $400M in ChatGPT rival Anthropic.
When will that happen? In years to come, once the hardware can keep up with the algorithm on a huge dataset such as "the whole web.:
Until then, good luck to anybody thinking they can use GPT-3 to build a useful search engine containing all the data available on the web.
